腾讯混元开源Hy4 flash模型,支持百万token上下文
腾讯混元11月18日宣布开源Hy4 flash模型,包含Hy4 flash、Hy4 flash-FP8和Hy4 flash Base三个版本,开发者可在GitHub、Hugging Face、ModelScope和GitCode下载部署。
据腾讯混元介绍,Hy4 flash总参数量224B,其中包含64B Engram参数,每个token激活12B参数,支持多模态输入与100万token上下文长度。该模型被定位为迄今性价比最高的Hy模型,API价格较Hy4大幅降低。
Hy4 flash已在腾讯云TokenHub、WorkBuddy、元宝、企业微信、QQ、ima、Marvis等产品上线。腾讯混元同时预告,正在加速开源Hy4模型。
开源地址:https://github.com/Tencent-Hunyuan/Hy4-flash
- 2026-11-19 00:46 | X:We’re hosting an AMA on Reddit this Friday, Nov 20, 8–10 AM PT! 🚀 Fresh off shipping Hy4 & Hy4 flash, our research team & @ShunyuYao12 will hang ...阅读原文
🚀 Fresh off shipping Hy4 & Hy4 flash, our research team & @ShunyuYao12 will hang out to unpack how we built our ultra-fast, lowest-cost everyday model.Tencent Hy: Hy4 flash is now open-weight, including Hy4 flash-FP8 and Hy4 flash-Base.224B (12B active), 1M context, natively multimodal.Fast, effic...
- 2026-11-18 17:00 | X:RT Ash Hart: I’ve been testing this with early access and I can confirm, it’s really good.阅读原文
Tencent Hy: Hy4 flash is now open-weight, including Hy4 flash-FP8 and Hy4 flash-Base.224B (12B active), 1M context, natively multimodal.Fast, efficient and built for everyday productivity.It's yours now. And stay tuned for Hy4 coming up next!🤗: https://huggingface.co/tencent/Hy4-flash🐙: https://gi...
- 2026-11-18 15:45 | X:RT seteiro: hy4-flash 正式开源啦,有需要的朋友可以去下载了! 说实话我感觉这个模型还是很不错的,可以适配于低烈度的需求了,大家可以去下载一手 模型下载地...阅读原文
RT seteirohy4-flash 正式开源啦,有需要的朋友可以去下载了!说实话我感觉这个模型还是很不错的,可以适配于低烈度的需求了,大家可以去下载一手模型下载地址GitHub:http://github.com/Tencent-Hunyua…HuggingFace:http://huggingface.co/tencent/Hy4-fl…Modelscope:http://modelscope.cn/models/Tencent…GitCode:http://ai.gitcode.com/tencent_hunyua…GGUF:http://huggingface.co/AngelSli...
- 2026-11-18 15:28 | X:RT ModelScope: Tencent Hy has released Hy4-flash as an open-weight 220B MoE model that activates just 12B parameters per token, with native image unde...阅读原文
RT ModelScopeTencent Hy has released Hy4-flash as an open-weight 220B MoE model that activates just 12B parameters per token, with native image understanding and 1M context.🤖 https://modelscope.ai/models/Tencent-Hunyuan/Hy4-flash🏆 Built for lower-cost agent work: On Tencent’s internal WorkIndex, H...
- 2026-11-18 14:34 | 界面新闻:沐曦股份曦云C系列GPU完成腾讯混元Hy4 flash适配阅读原文
11月18日,腾讯混元宣布正式开放Hy4 flash大模型权重。随后,沐曦股份表示,曦云C系列GPU完成Hy4 flash Day 0全链路适配,模型当日即在曦云C系列GPU完成推理验证,支持高吞吐、低成本规模化Agent业务落地。
- 2026-11-18 14:01 | X:RT 𝗸𝘂𝘀𝘂𝗿𝗶.: Qwen4より先にHy4の方が先に公開されてる!?!? 総パラメータ224Bでアクティブ12Bだから大体FNの倍くらいだしQwen3.8 Flash Ne...阅读原文
Tencent Hy: Hy4 flash is now open-weight, including Hy4 flash-FP8 and Hy4 flash-Base.224B (12B active), 1M context, natively multimodal.Fast, efficient and built for everyday productivity.It's yours now. And stay tuned for Hy4 coming up next!🤗: https://huggingface.co/tencent/Hy4-flash🐙: https://gi...
- 2026-11-18 13:26 | X:RT Ivan Fioravanti: Hy4 Flash weights are here! Base model included! 🔥阅读原文
Tencent Hy: Hy4 flash is now open-weight, including Hy4 flash-FP8 and Hy4 flash-Base.224B (12B active), 1M context, natively multimodal.Fast, efficient and built for everyday productivity.It's yours now. And stay tuned for Hy4 coming up next!🤗: https://huggingface.co/tencent/Hy4-flash🐙: https://gi...
- 2026-11-18 13:00 | X:RT vLLM: 🎉 Day-0 support for @TencentHunyuan’s Hy4 flash is here in vLLM! A 1M-token context and multimodal understanding for research, documents,...阅读原文
RT vLLM🎉 Day-0 support for @TencentHunyuan’s Hy4 flash is here in vLLM!A 1M-token context and multimodal understanding for research, documents, slides and computer use.224B total parameters, including 64B Engram, with 12B active. Hybrid attention, Latent MoE and Engram put efficiency at the center ...
- 2026-11-18 12:36 | 界面新闻:腾讯混元开源Hy4 flash阅读原文
11月18日,腾讯混元宣布开源Hy4 flash,包含Hy4 flash、Hy4 flash-FP8和Hy4 flashBase,开发者可在Git Hub、Hugging Face、Model Scope和Git Code下载部署。此外,公司也在加速开源Hy4模型。
- 2026-11-18 12:36 | IT之家:腾讯混元 Hy4 flash 模型官宣开源阅读原文
IT之家 11 月 18 日消息,腾讯混元今日宣布开源 Hy4 flash 模型,包含 Hy4 flash、Hy4 flash-FP8 和 Hy4 flash Base,开发者可在 GitHub、 Hugging Face、ModelScope 和 GitCode 下载部署。腾讯混元官方预告,正在加速开源 Hy4 模型,敬请期待。Hy4 flash 定位迄今性价比最高的 Hy 模型,总参数量 224B,其中包含 64B Engram 参数,每个 token 激活 12B 参数,支持多模态输入与 100 万 token 上下文长度,API 价格较 Hy4 大幅降低。Hy4 flash 已在腾讯云...